Continued-As-New workflow waits 1 minute before start

[bulat]

Continued-As-New workflow waits 1 minute before start if it receives UnhandledCommand:

UnhandledCommand in previous workflow:

Signal channel drain:

Workflow output:

Can i get rid of this delay?

[tihomir]

Could you provide the JSON logs (via tctl) rather than screenshots for the two executions please?

tctl wf show -w <wfid> -r <runid> --output_filename myhistory.json

Regarding UnhandledCommand, this is I believe expected here as your workflow code is trying to call ContinueAsNew, and signals are still being processed. In this case UnhandledCommand allows these signals to finish processing and not be lost before the execution is completed and a new one created via ContinueAsNew.

Do you have SDK metrics set up?

[bulat]

Logs: logs - Google Drive

Do you have SDK metrics set up?

How to check that?

[bulat]

After change to default docker-compose (docker-compose/docker-compose.yml at main · temporalio/docker-compose · GitHub) delay is gone.

Before that i used cassandra build docker-compose/docker-compose-cass-es.yml at main · temporalio/docker-compose · GitHub

[bulat]

After several checks I got 1 minute delay in another workflow, event id 171:

3.json

[bulat]

So how do I solve this problem?